Purpose
Segments lapsed customers — those who placed at least one order but have not purchased again within a configurable window — and tags them for re-engagement. This skill handles the Shopify-native data layer; sending re-engagement emails requires an external tool.
Prerequisites
- Authenticated Shopify CLI session:
shopify auth login --store <domain> - API scopes:
read_customers,write_customers
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
| store | string | yes | — | Store domain |
| format | string | no | human | human or json |
| dry_run | bool | no | false | Preview without tagging |
| inactive_days | integer | no | 90 | Days since last order to qualify as lapsed |
| min_orders | integer | no | 1 | Minimum lifetime order count to include |
| tag | string | no | win-back | Tag applied to lapsed customers |
| max_customers | integer | no | 500 | Maximum customers to process per run |
Workflow Steps
OPERATION:
customers— query Inputs: filterlast_order_date:<(NOW - inactive_days days),orders_count:>=(min_orders),first: 250, pagination Expected output: List of customer objects withid,defaultEmailAddress { emailAddress },firstName,lastName,ordersCount,lastOrder.processedAt; paginate untilhasNextPage: falseOPERATION:
tagsAdd— mutation Inputs: Customerid, tag string fromtagparameter Expected output: Confirmation per customer; collectuserErrors
GraphQL Operations
# customers:query — validated against api_version 2025-04
query LapsedCustomers($first: Int!, $after: String, $query: String) {
customers(first: $first, after: $after, query: $query) {
edges {
node {
id
defaultEmailAddress {
emailAddress
}
firstName
lastName
ordersCount
lastOrder {
processedAt
totalPriceSet {
shopMoney {
amount
currencyCode
}
}
}
}
}
pageInfo {
hasNextPage
endCursor
}
}
}
# tagsAdd:mutation — validated against api_version 2025-01
mutation TagsAdd($id: ID!, $tags: [String!]!) {
tagsAdd(id: $id, tags: $tags) {
node {
id
}
userErrors {
field
message
}
}
}
Session Tracking
Claude MUST emit the following output at each stage. This is mandatory.
On start, emit:
╔══════════════════════════════════════════════╗
║ SKILL: Customer Win-Back ║
║ Store: <store domain> ║
║ Started: <YYYY-MM-DD HH:MM UTC> ║
╚══════════════════════════════════════════════╝
After each step, emit:
[N/TOTAL] <QUERY|MUTATION> <OperationName>
→ Params: <brief summary>
→ Result: <count or outcome>
If dry_run: true, prefix mutation steps with [DRY RUN] and do not execute.
On completion, for format: human:
══════════════════════════════════════════════
OUTCOME SUMMARY
Lapsed customers found: <n>
Customers tagged: <n>
Errors: <n>
Output: winback_<date>.csv
══════════════════════════════════════════════
For format: json, emit the standard JSON schema with outcome keys: lapsed_found, customers_tagged, errors, output_file.
Output Format
CSV winback_<YYYY-MM-DD>.csv with columns:
customer_id, email, first_name, last_name, orders_count, last_order_date, tag_applied
Error Handling
| Error | Cause | Recovery |
|---|---|---|
THROTTLED |
Rate limit | Wait 2s, retry up to 3 times |
userErrors on tagsAdd |
Customer not found or invalid ID | Log, skip, continue |
Best Practices
- Use a dated tag (e.g.,
win-back-2026-04) so you can track which cohort was targeted each month and avoid re-tagging customers who already received a win-back campaign. - Set
min_orders: 2to focus on customers who had a genuine purchase relationship, not one-time buyers who may never have intended to return. - Run with
dry_run: truefirst to validate the lapsed customer count before tagging — the count informs the scale of your re-engagement campaign.